Develop a molecular-dynamics-based spin relaxation theory

Develop a molecular dynamics–based theory of spin–lattice relaxation that enables the estimation of spin–phonon relaxation times directly from time–correlation functions of fluctuations of spin Hamiltonian tensors (such as the zero–field splitting tensor D and crystal–field parameters B_m^l) computed along molecular dynamics trajectories, and extend the current second–order perturbative formalism to a fourth–order theory capable of describing Raman relaxation processes.

Background

The paper introduces a workflow that uses machine-learning force fields and equivariant models to predict spin Hamiltonian tensors and compute their time–correlation functions along long molecular dynamics trajectories. These correlation functions naturally incorporate non–equilibrium effects, phonon dissipation, and temperature–dependent linewidths beyond the equilibrium harmonic approximation.

While the authors can compute these correlation functions, they explicitly note that a proper molecular dynamics spin relaxation theory linking such correlations to quantitative relaxation times is currently unavailable. They also state that existing MD-based treatments are limited to second-order perturbation theory, whereas Raman relaxation requires a fourth-order formalism. Hence, establishing a rigorous MD-based theory is an outstanding problem necessary to exploit these correlation functions for predictive spin–lattice relaxation simulations.

References

The correlation functions could in principle be used to estimate the relaxation time. However, a proper molecular dynamics spin relaxation theory is not available yet, and its derivation is beyond the scope of this work. Indeed, whilst MD is able to include non-equilibrium effects and phonons' dissipation, it is based on a classical notion of nuclei, which is expected to be of limited accuracy at such low temperatures. Moreover, MD has only been adapted to second-order perturbation theory so far, whilst Raman relaxation will require the development of a fourth-order theory.

A machine-learning framework for accelerating spin-lattice relaxation simulations (2410.08912 - Briganti et al., 11 Oct 2024) in Results, subsection "Machine learning the full potential energy surface and its spin Hamiltonian" (paragraph beginning "The correlation functions could in principle be used to estimate the relaxation time.")